Machine Learning Explainability

Go to class
Write Review

Machine Learning Explainability provided by Kaggle is a comprehensive online course, which lasts for 4 hours worth of material. Machine Learning Explainability is taught by Dan Becker. Upon completion of the course, you can receive an e-certificate from Kaggle. The course is taught in Englishand is Free Certificate. Visit the course page at Kaggle for detailed price information.

Overview
  • Extract human-understandable insights from any model.

    • Why and when do you need insights?
    • What features does your model think are important?
    • How does each feature affect your predictions?
    • Understand individual predictions
    • Aggregate SHAP values for even more detailed model insights

Syllabus
    • Use Cases for Model Insights
    • Permutation Importance
    • Partial Plots
    • SHAP Values
    • Advanced Uses of SHAP Values